Risk Assessment for Vibrio vulnificus
Angelo DePaola & John Bowers
Presentation to ISSCAugust 11, 2004Las Vegas, NV
History of FAO/WHO Risk Assessments for Vibrio spp. in
Seafood� March 2001 – Initial meeting of drafting
group� CCFH requested Vp and Vv in raw shellfish� Vp in raw finfish (Japan) & Cholera in
warm water shrimp for export� 4 drafting group meetings and 2 expert
consultations� Peer review and publication in 2004
Risk Assessment
� Hazard identification� Exposure assessment� Hazard characterization� Risk characterization
Vv Hazard Identification
� Naturally occurring estuarine bacterium� Warm moderately saline waters� Three biotypes (1,2 &3)� Wound, gastroenteritis, primary septicemia� Preexisting chronic illness� 50% fatality rate� Raw Gulf Coast oysters
V. vulnificus in Gulf Coast OystersFactors supporting risk assessment
� Consistent high reporting for septicemia� Seasonal relationship with exposure & cases� Dominant vehicle of transmission� Shell storage prevents cross contamination� Raw consumption eliminates cook
variability and uncertainty
V. vulnificus in Gulf Coast OystersFactors supporting risk assessment
� Quantitative data on V. vulnificus levels at harvest and consumption
� Growth and survival of natural populations in oysters
� Availability of V. parahaemolyticus Risk Assessment on raw oysters
Objectives
� Adapt FDA-VPRA model to assess risk of Vv in raw oysters
� Identify most appropriate data/data gaps and limitations for modeling Vv in raw oysters
� Assumptions – grounded by related data� Conduct risk characterization of Vv in raw
oysters
� Evaluate targeted mitigation levels for risk reduction for Vv illness
Conceptual V. vulnificus Model
Vv/g (numbers) at consumption
Ingested dose
RISK OF
Weight per oyster (g)
Storage time
Oysters per serving
Susceptible population ILLNESS
Vv/g at cooldown
Cooldown time
POST-HARVEST
Water temperature
total Vv/g
Vv/g at harvest
Vv/g at 1strefrigeration
Air temperatureTime to refrigeration
HARVEST
Regional, seasonal & yearly variation
Vv/g (numbers) at consumption
Ingested dose
RISK OF
Weight per oyster (g)
Storage time
Oysters per serving
Susceptible population ILLNESS
Vv/g at cooldown
Cooldown time
POST-HARVEST
Water temperature
total Vv/g
Vv/g at harvest
Vv/g at 1strefrigeration
Air temperatureTime to refrigeration
HARVEST
Regional, seasonal & yearly variation
Needed Inputs
� Exposure assessment� Vv levels at harvest� % pathogenic� Vv growth rates� Vv survival rates
� Hazard characterization� Susceptible population� Dose response
Vv at harvest
� Motes et al. 1998� Weekly samples (July 1994-Sept. 1995)� FL, AL, LA, TX� MPN duplicate oyster (12) samples
� Temperature� Salinity
������������� ����� �������������������������������� ���������� ��
���
�
��
���
����
�����
������������ ���������
� � � � � � � � � � � �
�����
Effect of Temperature on Vv densities in Gulf Coast oysters
0.01
0.1
1
10
100
1000
10000
100000
1000000
40 50 60 70 80 90
water temperature (F)
Vv
dens
ity in
oys
ters
(MP
N/g
)
Motes (averaged replicates)regression fit (MLE)
Effects of high salinity on exposure and risk
� Salinity not included in VVRA� Effects minimal for typical Gulf Coast
salinities (10-30 ppt) � Vv levels low or nondetectable in NC & SC
sites with high salinity (>30 ppt) & temp.� High salinity areas typical in Europe, Asia,
Australia and New Zealand
Effect of salinty >30 ppt in US oysters
8.53199All
4.23614>30°C (86°F)
2.7233025-30°C (77-86°F)
19.5303320-25°C (68-77°F)
2.84122<20°C (68°F)
Vv/g% Vv detectable
Number of samples
Temperature range
Vv growth and survival in oysters
Study Holdingtemperature(Celsius)
Growth rate(log10 per hr)
Assumptions/Limitations
Cook, 1997 28 0.175 Ambient air temperature variedfrom 24 -33, assumed average of 28oC
Cook, 1994 18 0.025 Rate per hour assumed constantwith observed average 0.75 logincrease (n=5) over period of 30hours
Kaspar and Tamplin, 1993 13 Presumed no growth temperature
Cook et al. 2002 5.7 -0.002 Range 0-16°C representative ofoyster industry cooler temperatures
Vv/g exposure predictions
� Summer� log Vv/g at harvest:
3.27 (0.64)� log Vv/g at 1st
refrigeration:4.00 (0.74)
� log Vv/g after cooldown:4.46 (0.77)
� log Vv/g at consumption:4.15 (0.78)
� Winter� log Vv/g at harvest:
0.47 (1.09)� log Vv/g at 1st
refrigeration:0.57 (1.16)
� log Vv/g after cooldown:0.63 (1.21)
� log Vv/g at consumption:0.30 (1.22)
Model Validation
� Validation of model predictions against data not used in model construction
� Data available for this is Vv at consumption (retail study)
May '99
�� ��������� ���!�"�#���#���#�$�#���#���#���#���#���
���������� ����
Predicted and observed levels of Vv
Hazard characterization parameters
� Mean water temperature� DI buoy 1987-97 monthly avg.
� Servings for at risk individuals� 50% NMFS landings consumed raw� 7% of population at risk
� Mean Vv/serving� Model prediction based on water & air temp.� Mean serving size of 196g
� Vv cases � Reported primary septicemia cases
Vv risk factors in U.S.Risk factor Prevalence per 100,000 individuals
Diabetes (insulin-dependent) 540.5
Liver disease (cirrhosis) 2000.0 (range: 1600 - 9900)
Gastric acidity 38.9
Cancer 1420.0
Hepatitis (B and C) (range: 400 - 1600)
Kidney disease 108.0
Haemochromatosis 1081.1
AIDS
Immune-compromised due totreatment/surgery
540.5
Asthma 25.7
Rheumatoid arthritis 51.4
Psoriatic arthritis 37.9
Lupus (range: 4 - 250)
Polymylagia rheumatica 53.0
Giant cell arthritis 12.0
Transplant recipients 59.5
Dose response obstacles� Cases rare among at risk population
� variability in strain virulence� variability in susceptibility of population
� Animal models not reliable� Lack of agreement between studies� Route of administration (oyster
consumption)
� Controlled human volunteer studies unethical
Shellfish Associated V. vulnificus Illnesses 1995 through 2001
0
5
10
15
20
25
30
35
40
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month
Model inputs and Vv casesMonth
Mean and Std dev of water
temperature Servings for at risk
individuals Mean V. vulnificus per
serving (dose) Average #
Cases
Jan 12.9 (2.9) 128,000 14,000 0.14 Feb 15.1 (2.8) 132,000 70,000 0.14 Mar 17.4 (2.0) 151,000 109,000 0.29 Apr 21.7 (1.7) 131,000 675,000 1.86 May 25.8 (1.9) 110,000 5,025,000 4.57 Jun 28.8 (1.4) 105,000 11,561,000 3.14 Jul 30.0 (1.2) 97,000 15,598,000 4.14
Aug 30.3 (1.0) 88,000 16,536,000 5.14 Sep 28.2 (1.7) 99,000 9,008,000 4.71 Oct 22.7 (2.7) 127,000 1,943,000 4.43 Nov 18.4 (2.8) 146,000 257,000 2.71 Dec 15.4 (2.5) 149,000 39,000 0.71
Beta Poissson vs Exponential
Predicted Cases by Beta Poisson and Exponential vs Observed Cases
7.851.608.03Fall
13.9916.6012.28Summer
9.577.5911.86Spring
0.570.030.52Winter
Observed cases
Exponential cases
Beta Poisson
cases
Season
Post Harvest Processing
� Approved or proposed technologies� mild heat treatment� freezing� irradiation� hydrostatic pressure
� Reduce V.v. to non-detectable (<3 MPN/g)� HACCP plan � Label: “Processed to reduce V.v. to non-detectable
levels”
Annual Vv illnesses at targeted levels of mitigation
7.7 (3.8, 15.3)5.26 x 10-6 (2.60 x 10-6, 1.05 x 10-6)300/g
1.2 (0.5, 3.1)8.20 x 10-7 (3.42 x 10-7, 2.12 x 10-6)30/g
0.16 (0.06, 0.4)1.09 x 10-7 (4.10 x 10-8, 2.73 x 10-7)3/g
Annual number of cases (mean and 95% uncertainty interval) a
Risk per serving(mean and 95%
uncertainty interval)
Target
Scenario analysis by RA� Time/temperature controls
� ISSC time/temperature matrix for Vv 1997� Canadian immediate cooling for Vp 2000
� Regions or countries: different ecology or practices than Gulf Coast� High salinities for Vp & Vv (New Zealand)� Intertidal harvest in Pacific NW
� Mitigations for other pathogens� Warm temperature depuration of Norwalk
in UK
Effect of time unrefrigerated on numbers of Vv cases
19.28 (16.11, 24.06)20 hr
15.48 (13.49, 18.82)10 hr
11.59 (9.78, 14.08)5 hr
6.77 (5.27, 8.45)0 hrSpring
5.12 (1.29, 11.05)20 hr
1.08 (0.23, 4.45)10 hr
0.40 (0.09, 1.96)5 hr
0.19 (0.06, 0.68)0 hrWinter
Expected # cases(90% uncertainty
range)
Timeunrefrigerated
Season Expected # cases(90% uncertainty
range)
Time unrefrigerated
Season
17.30 (13.72, 21.98)20 hr
11.64 (8.91, 15.72)10 hr
7.37 (4.66, 10.62)5 hr
3.06 (1.64, 5.46)0 hrFall
17.55 (15.51, 21.66)20 hr
15.31 (12.93, 18.34)10 hr
12.16 (10.46, 14.04)5 hr
7.65 (6.57, 8,82)0 hrSumme
r
��������������� ��������������� � � �� ��� � �� � ��� � � �� � �� �� �� �� �
12.2 (10.5, 14.1)
4.12 (0.78)5.13 (2.77)Post 1996 changeSummer
12.9 (11.1, 15.0)
4.25 (0.81)5.27 (2.81)Pre 1996 changeSummer
11.7 (9.8, 14.0)3.27 (1.08)5.13 (2.77)Post 1996 changeSpring
12.6 (10.7, 15.2)
3.36 (1.11)5.27 (2.81)Pre 1996 changeSpring
Expected # cases
Mean and std dev of Log
Vv/g at consumption
Mean and std dev of time
unrefrigerated
Time unrefrigerated
Season
Effect of salinty >30 ppt on Vv level and risk
(1.3x10-8, 6.9 10-7)4.2>30°C (86°F)
(8.3x10-9, 4.6 10-7)2.725-30°C(77-86°F)
(6.5x10-8, 2.9 10-6)19.520-25°C(68-77°F)
(1.0x10-8, 3.4 10-7)2.8<20°C (68°F)
Risk/serving(range-best & worst case for post harvest
growth)
Vv/gHarvest
Temperature range
Future Directions
Remote Sensing
May 4, 2004 SST
May 4, 2004 SSTZoom of MS, AL Coast
Model Equations for the Gulf
� mean(log(Vp/g)) = -0.63 + 0.12*WTEMP
� mean(log(risk)) = -7.23 + 0.14*WTEMP
� Approximation of VPRA formulas
� Other formulas (i.e. mitigations) possible
May 4, 2004 Mean Log Vp/gZoom in of MS, AL coast
May 4, 2004 Mean Risk
Advantages of remote sensing
� Surgical management of risk instead of using sporadic distribution NDBC
� Real time posting of risk on website� Predictions of mitigation or PHT times� Potential to refine VPRA or VVRA using
remote sensing data other than SST� Objective measurements for international
harmonization
Conclusions
� VPRA provided suitable framework & many parameters transferable to VVRA
� Temperature based predictions of exposure validated by market data
� Beta poisson fit for dose response agrees with seasonality of Vv cases
� Interventions to reduce Vv illnesses can be evaluated with confidence
� Remote sensing may provide real time objective measurements of risk & facilitate international harmonization